78results about How to "Reduce Data Transfer Latency" patented technology

The invention discloses a Transform large model training method based on cloud edge collaboration. The method comprises the following steps: S1, constructing a data set; s2, preprocessing the data set; s3, segmenting the preprocessed data set by using a data parallelism strategy, and performing distributed training by combining data parallelism and model parallelism technologies; s4, monitoring computing resources and network conditions of the edge equipment and the cloud server in real time, dynamically adjusting a data distribution strategy, and optimizing computing load distribution; s5, a multi-agent dominant actor-commentator algorithm is combined with a multi-arm tiger machine strategy, and the layered structure of the Transform model is adjusted in real time; s6, performing partial model reasoning tasks on the edge equipment in combination with the edge reasoning capability; s7, protecting data privacy by adopting a federal learning strategy and utilizing a differential privacy technology; and S8, performing hardware acceleration by utilizing heterogeneous computing resources. According to the method, a cloud edge cooperative computing and intelligent scheduling method is adopted, and large model training and reasoning with efficient privacy protection are realized.

The invention relates to the technical field of force feedback handles, in particular to a remote surgical robot control system based on a force feedback handle and a control method thereof. The method comprises the following steps: carrying out multi-dimensional force data acquisition on an end effector of the surgical robot, and carrying out contact force size, direction and time sequence feature extraction to obtain multi-dimensional force feature data; performing surgical operation safety baseline analysis according to the multi-dimensional force feature data to generate surgical operation safety baseline data; performing dynamic force feedback threshold calculation according to the surgical safety operation baseline data to obtain force feedback threshold range data; and performing adaptive compression processing on different force feedback signals according to the force feedback threshold range data to obtain dynamic force feedback control parameters. Through real-time control instruction optimization and a closed-loop feedback strategy, the system can quickly respond to an operation intention, early warning is carried out on potential risks, and it is ensured that the operation process is safer and more reliable.

The invention relates to the technical field of offshore wind power operation risk analysis, and discloses an offshore wind power operation risk analysis system based on deep learning, and the system comprises a data collection and preprocessing module which is used for collecting wind power equipment operation state data and environment data of multiple types of sensors, and carrying out the cleaning and preprocessing of the data; the multi-modal data fusion and feature extraction module is used for extracting and fusing multi-modal features from different data sources; and the real-time risk prediction and intelligent decision module is used for analyzing the operation state of the wind power equipment in real time based on a deep learning model, and performing prediction and alarm when a fault trend is detected. By introducing a self-attention mechanism and a Transform architecture, deep fusion analysis is performed on multi-modal data from wind power equipment, and information of different data sources can be weighted and integrated in a unified feature space, so that the system can adaptively focus on important features, and the analysis precision and fault detection capability of the multi-modal data are remarkably improved.

The invention provides a parallel water pump intelligent algorithm optimization combination and control method, which comprises the following steps of: configuring a sensor and a calculation module on an edge node of each water pump, acquiring and preprocessing operation data of each water pump in real time, and generating a unified real-time feature vector and a time sequence feature; dynamically calculating the service life state of the water pump according to the real-time feature vector, and quantifying the residual service life; designing a dynamic start-stop scheme based on a fuzzy logic algorithm according to the service life state, and balancing load requirements and water pump service life distribution; and based on federal learning, the residual life of the water pump is optimized in a global perspective, and an operation strategy based on a start-stop scheme is dynamically optimized and adjusted. A complete solution is formed, and a brand new technical path is provided for efficient and reliable operation of the parallel water pumps.

The invention relates to the technical field of unmanned aerial vehicles, in particular to a limited space cable unmanned aerial vehicle inspection system based on a GPS signal-free condition, which comprises an unmanned aerial vehicle navigation module, a cable detection module, an obstacle avoidance module, an environment monitoring module, a data processing and recording module and a power management module, autonomous navigation of the unmanned aerial vehicle in a limited space is realized, the cable detection module monitors the surface state of a cable based on a convolutional neural network model, the obstacle avoidance module dynamically detects obstacles and adjusts a flight path, and the environment monitoring module monitors temperature, humidity and air quality and gives an early warning when the temperature, humidity and air quality exceed a safety threshold. The data processing module processes data in real time in combination with an edge computing technology and uploads the data to a control center, and the power management module regulates and controls energy consumption through an intelligent algorithm and automatically returns or emergently lands when the electric quantity is insufficient; the method is suitable for a cable inspection task in a GPS-signal-free environment.

The invention relates to the technical field of smart home, and discloses an AI-based fan intelligent temperature regulation control system, which comprises a multi-dimensional data acquisition module for acquiring environment temperature, humidity, air quality, illumination intensity and human body state data; the data fusion and preprocessing module is used for carrying out denoising, standardization and time synchronization on the multi-dimensional data and fusing the multi-source data; the AI decision and control module is used for generating a speed regulation and mode control instruction of the fan based on the human body comfort evaluation model and an AI algorithm; and the feedback optimization and execution module is used for executing the control instruction. Through a multi-dimensional data perception and fusion technology, comprehensive monitoring of environment temperature and humidity, air quality and a human body state is realized. The AI intelligent regulation and control algorithm is combined, the fan speed and mode are dynamically adjusted, and the user comfort and energy consumption are balanced; through edge calculation and cloud optimization, real-time response and continuous adaptive optimization of the system are ensured, and the accuracy and stability of the system are improved.

The invention discloses an inspection robot charging control system and a charging device thereof, relates to the technical field of inspection robots, and realizes the optimal balance of the inspection robot between the charging demand and task execution through a dynamic task and a charging scheduling algorithm. Data such as the position, the electric quantity, the current task state and the environment condition of the robot are monitored in real time; edge computing nodes are deployed in an inspection area, sensor data are processed and analyzed in real time, the burden of a central server is relieved, and data transmission delay is reduced; based on the characteristics of the inspection task and the movement track of the robot, charging points are flexibly arranged in the inspection area; through Internet of Things data analysis, the positions and the number of charging points are dynamically adjusted, and the return problem caused by fixed charging points is avoided; the mobile charging station actively goes to the position of the robot for charging according to the position and the electric quantity state of the robot, and the return time of the robot is shortened.

The invention discloses an Internet of Vehicles driving state monitoring method and system, and the method comprises the steps: 1, collecting the driving state data in real time, and enabling the driving state data to comprise the behavior data of a driver and environment data; step 2, preprocessing the collected multi-source heterogeneous data at a vehicle-mounted edge computing node; step 3, inputting the fused data into a variational auto-encoder, and when the variational auto-encoder extracts the low-dimensional hidden variable representation of the driving state data, deducing hidden variable distribution by optimizing a variational lower bound; 4, inputting the inferred hidden variables into the fractional order differential equation model to obtain a predicted value of the driving state; step 5, federated learning is carried out on a plurality of vehicle-mounted edge nodes, and collaborative optimization of a global model is realized while data privacy is protected; according to the method, distributed collaborative decision-making of multiple agents is realized by solving Nash equilibrium, the adaptability and robustness of the system are improved, and the traffic safety and the driving experience can be remarkably improved.

The invention relates to the technical field of safety engineering, in particular to an intelligent high-altitude and fire operation safety management system based on risk assessment, which comprises a safety helmet module, a data acquisition and transmission module, an edge calculation and processing module, a risk assessment and decision module, an emergency response module and a data analysis and optimization module. The intelligent safety helmet worn by each operator is integrated with a flame sensor and a height sensor, and flame information in an operation environment and height changes of the operators can be collected in real time. Through the low-power-consumption wireless communication technology, the collected data is transmitted to the edge computing device in real time, and the edge computing device calculates the comprehensive risk index through multi-sensor data fusion. And when the risk index exceeds a preset threshold value, the system automatically triggers alarm, sends early warning information to operators and a management platform, and takes emergency measures, so that the accident rate in high-altitude operation and fire operation can be effectively reduced, and the safety management level of the operation environment is improved.

The invention discloses a link bandwidth allocation method and system for multilink aggregation of a multimodal mobile network. The method comprises the following steps: acquiring a mobile network base station, an equipment position and real-time environment data; an intelligent power control algorithm is adopted, and transmission power is dynamically adjusted according to environment data and network loads; acquiring real-time signal intensity data and historical data from the network equipment to the base station; predicting a minimum transmission point of signal transmission based on real-time and historical data; adjusting the direction and position of an antenna according to the position of the base station, and designing connection by adopting a beam forming antenna; setting a threshold value according to the real-time weather and the minimum transmission point data, and deploying a signal enhancer; according to the link bandwidth allocation method and system for multi-link aggregation of the multi-modal mobile network, the link quality is monitored in real time, the network is dynamically adjusted, the network load is balanced through an intelligent algorithm, the signal transmission environment is evaluated in combination with real-time weather data, the signal enhancer is deployed, and the signal coverage range is expanded.

The invention belongs to the technical field of energy conservation and emission reduction, and particularly relates to a data center low-load air conditioner energy conservation control method and system.The method comprises the steps that an upper computer monitors an interface in real time to conduct visual digital modeling on a data center through a data center digital modeling method, and the data center is displayed in a 3D simulation and digital mode in real time; the network data communication system realizes data interaction and real-time communication by using a network communication link construction method; the data sensing system obtains equipment operation data in real time and supports energy-saving control of the air conditioner; the temperature prediction system is applied to the temperature change prediction method, and preprocessing of collected data, data set manufacturing and training of a neural network model are achieved. The equipment parameter control system is internally provided with an equipment operation parameter optimization control method, optimal combination optimization of air conditioner internal equipment operation parameters is achieved, and an instruction is issued and transmitted to the console to achieve control. And the console adjusts and controls the related parameters of the air conditioning equipment according to the control instruction received from the network data communication system.

The invention provides a data acquisition and monitoring analysis system based on an MES platform, and relates to an industrial software application system. Comprising a low-code development module, a factory modeling module, a monitoring module and a data acquisition module, the low-code development module realizes cross-device linkage development and control program generation through combination of a dragging interface and a device API library in an MES platform integration environment, and guarantees efficient communication control of the device, the factory modeling module integrates two production modes, and the monitoring module is connected with the monitoring module. The monitoring module is used for realizing dynamic modeling and optimizing a production process according to multi-dimensional parameters, the monitoring module is used for realizing ultra-low-delay transmission processing and real-time monitoring of data under high concurrency by applying a dual-cache and message middleware mechanism, and the data acquisition module is used for realizing real-time monitoring of the data based on a machine learning algorithm and a complex event processing technology. Dynamically adjusting the acquisition frequency precision based on the production state; the system aims at improving the efficiency, quality and management level of industrial production, and has wide application prospects and important industrial software application value.

The invention relates to the technical field of equipment fault detection, and discloses an equipment fault prediction and diagnosis method which is applied to an ATE system, and the method comprises the steps: collecting the real-time operation data of equipment through a plurality of sensors, and dynamically scheduling the sensors through an adaptive algorithm, so as to adjust the data collection frequency and range according to the state of the equipment; the collected data is preprocessed; training a plurality of machine learning models based on the historical fault data and the operation data, and deploying the models to an ATE system for real-time analysis; predicting an equipment fault according to the model, generating alarm information when a potential fault is detected, and sending the alarm information to an operator through a user interface or a notification system; generating a fault diagnosis report according to an analysis result and providing a repair suggestion; recording a fault instance and data, and updating and retraining the machine learning model to improve prediction and diagnosis accuracy; and generating a digital twin model of the equipment based on real-time and historical data, simulating the running state of the equipment, predicting future faults, and evaluating the health condition of the equipment.

The invention discloses an industrial scene-oriented intelligent Agent system and an implementation method thereof, and the system comprises a user interaction layer which is used for providing a diversified interface and carrying out the communication with a back-end service through an API gateway; the intelligent layer is used for realizing task management, task planning and resource scheduling functions by adopting a micro-service architecture arranged based on a container; the physical equipment layer is used for being integrated with an upper-layer system through an industrial Internet of Things protocol and processing time-sensitive applications by adopting an edge computing technology; wherein the user interaction layer is bidirectionally connected with the body intelligent layer, and the body intelligent layer is bidirectionally connected with the physical equipment layer. According to the invention, a series of operations such as order analysis, material planning, production planning, task distribution, material preparation, production execution, quality control and exception handling can be autonomously carried out, manual intervention is greatly reduced, and the production efficiency is improved.

The invention discloses a patient health data management system and method based on edge computing, and relates to the technical field of edge computing, and the system comprises an edge computing node, a cloud server, a data transmission module and a system optimization module. Comprising the following steps that 1, medical equipment transmits health data of a patient to an edge computing node in real time through a data acquisition module; 2, the data acquisition module sends the data to a data processing module, a deep learning algorithm is used for analyzing data features, and various formulas are applied for processing and optimization; 3, the processed data are encrypted and stored through a data storage and encryption module, and integrity verification is carried out regularly; 4, the edge computing node transmits required data to a cloud server, and a data transmission module is responsible for dynamically adjusting a transmission strategy according to network conditions; and 5, the cloud server analyzes and mines the multi-source data, and outputs feedback information to further optimize a processing algorithm and model parameters of the edge computing node.

The invention discloses a mass resource access edge method suitable for a gridding power grid, and relates to the technical field of power grids, and the method comprises the steps: dividing the gridding power grid into a plurality of hierarchical nested sub-grids, and configuring an edge calculation node for each sub-grid; acquiring an original resource data set in the sub-grids through edge computing nodes; analyzing the original resource data set by using a multi-factor weighted scoring algorithm to generate a second resource access priority list; obtaining a resource access scheme of each sub-grid according to the second resource access priority list and the real-time load state of the sub-grid; and constructing a distributed collaborative decision-making mechanism, and coordinating resource access schemes of adjacent levels of sub-grids to complete massive resource access. According to the method, the power grid is gridded and divided into the multi-level nested sub-grids, so that the resource access pressure of each region is effectively dispersed, and efficient access and management of resources are realized through optimized deployment of the edge computing nodes.

The invention provides a multi-agent cooperative edge caching method and system based on a gating communication mechanism, and the method comprises the steps: deploying agents in a plurality of edge servers, judging the caching states of the edge servers through employing a gating model, and determining whether to upload the state information of the corresponding agents to a central server or not according to a judgment result. After receiving the agent state information, the central server comprehensively considers the state information of each agent based on a multi-agent decision model, determines an optimal cache decision through a reinforcement learning algorithm, issues the optimal cache decision to the corresponding edge server, and receives the optimal cache decision by the edge server; and transmitting the file data to the client, and performing cache management according to the cache decision. According to the invention, the optimal position of the cooperative edge server can be automatically determined, the servers can selectively exchange information, and the waste of bandwidth resources is reduced.

The invention discloses a Zynq-based convolutional neural network hardware acceleration system, which is realized based on a Zynq heterogeneous chip and comprises a PL end module, a PS end module and a storage module. The PL end module comprises a convolution acceleration module, a full connection layer acceleration module and a Softmax activation module and is used for completing convolution calculation, full connection operation and activation function processing; the PS end module comprises an input data processing module and a post-processing module and is responsible for loading and preprocessing input data and sorting and screening detection results. The storage module is composed of an on-chip memory and an off-chip memory, wherein the on-chip memory stores an intermediate processing result of input data, a convolution kernel and intermediate feature map data; and the off-chip memory stores the original image data, the weight parameter of the full connection layer and the final detection result. All the modules are connected through a 64-bit AXI bus, the PS end module is responsible for coordinating data flow, and the PL end module executes efficient parallel computing. The system makes full use of the high-performance characteristic of the Zynq heterogeneous architecture, is suitable for an efficient reasoning acceleration scene of the convolutional neural network, and particularly has a wide application prospect in embedded AI equipment.

A virtual network resource allocation method in a multi-server scene is based on a network communication system of a virtual network and a physical network, resources needed by virtual nodes comprise computing power, memory and storage, and the demand of a virtual link is bandwidth; the physical network is composed of physical nodes and physical links, and the core of virtual network resource mapping is that virtual nodes and links are reasonably distributed to the physical nodes and the links; in a multi-server environment, communication among virtual nodes can cross a plurality of physical servers, so that the number and efficiency of cross-host links become the key to influence the system performance; a mapping algorithm based on the adaptation degree is adopted; judging whether the physical node can meet the requirement of the virtual node or not by calculating the adaptation degree between the virtual node and the physical node; the adaptation degree is calculated according to resource requirements of the virtual nodes and available resources of the physical nodes, and only the physical nodes with the adaptation degree reaching a set threshold value are regarded as effective mapping targets.

The invention discloses a cache-assisted satellite task unloading and resource allocation method, and the key points of the technical scheme are that the method comprises the following steps: S1, building a system model which comprises the steps: building a user unloading model in a satellite-ground fusion network, building an energy consumption model, and building a cache utility model; s2, establishing an optimization problem, taking minimization of system cache utility as a target function, and defining constraint conditions; s3, converting the defined optimization problem into a Markov tuple, wherein the Markov tuple comprises a state space, an action space and a reward function; s4, a method based on deep reinforcement learning is provided for training, and a final unloading strategy is obtained; according to the method, computing resources in the low-orbit satellite network are effectively utilized by introducing the cache technology, the task unloading process is optimized, dependence on a satellite communication link can be relieved, data transmission delay is reduced, the response speed and stability of the system are improved, and the unloading efficiency of the computing edge system in the satellite-ground fusion network is improved.

The invention relates to the technical field of smart home interconnection, in particular to a signal interconnection optimization method and system for smart home network equipment. The method comprises the following steps: identifying all intelligent network equipment nodes in a home area, and obtaining real-time signal states of all intelligent home network equipment; carrying out inter-device topological dependency mining on the intelligent network device nodes, and constructing a network device topological dependency graph; carrying out inter-device signal interference analysis on a network device topology dependence graph based on the real-time signal state so as to obtain signal interference characteristics in the region; performing signal transmission frequency calculation on the intelligent network equipment node to obtain the signal transmission frequency of the network equipment; and performing dynamic frequency coordination optimization on the signal transmission frequency of the network equipment based on the signal interference characteristics in the region, and constructing a dynamic frequency optimization strategy. According to the invention, the signal interconnection quality and efficiency of the smart home are improved.

The present invention relates to the technical field of electronic digital data processing, and specifically discloses a hierarchical federated learning method of adaptive compression coding, the method comprising: hierarchical federated learning architecture construction, edge server matching, model generation performance judgment, compression rate matching and hierarchical federated learning architecture judgment, firstly, a hierarchical federated learning architecture is constructed so that each layer can make full use of its computing resources, an edge server corresponding to a terminal device is obtained by matching, each terminal device is used to generate an environmental data training model, a corresponding compression rate is obtained according to the network transmission performance of the edge server, the edge server receives the environmental data training model of each terminal device and performs adaptive compression coding and edge aggregation in sequence according to the corresponding compression rate, a cloud server receives the edge aggregation result and performs global aggregation, and finally judges whether the cloud server to which the hierarchical federated learning architecture belongs uploads the corresponding result of the global aggregation to an environmental resource scheduling cloud platform.

The invention discloses a panoramic monitoring method for power transmission and transformation equipment, and belongs to the field of power system automation. A panoramic monitoring method for power transmission and transformation equipment comprises the following steps: S1, performing digital modeling of monitoring information on primary and secondary equipment in a power system, including equipment state, performance parameters and environmental factors; s2, carrying out graphical representation on the equipment state by utilizing the digital information model, and carrying out interactive operation through a user interface; s3, establishing information association between the devices, forming a comprehensive monitoring network of the power grid, and optimizing the accuracy of association between the devices by using a machine learning algorithm; s4, updating and displaying the monitoring state of the power grid in real time, including a geographical wiring diagram and an equipment state diagram, and identifying an abnormal mode through a data mining technology; and S5, utilizing a predictive maintenance algorithm to predict potential faults and recommend a maintenance plan according to the equipment state.

The invention relates to the field of micro-services, in particular to a cloud platform-based registration center switching method, which comprises the following steps of: establishing a plurality of transmission channels, and transmitting initial registration information through the plurality of transmission channels; receiving a plurality of addressing information and extracting feature addressing information; initial configuration parameters are set, and the feature addressing information is sent to the latest registration center to search target registration information; if the target registration information does not exist in the latest registration center, checking and repairing the transmission channel, transmitting the initial registration information again, and searching again after a first time period; if the re-searching result is that the target registration information does not exist in the latest registration center, the configuration parameters are replaced to send the feature addressing information to the initial registration center; and searching the target addressing information in the initial registration center, and transmitting a searching result to the newest registration center through the transmission channel to complete updating of the newest registration center. According to the invention, the stability of registration center switching is improved.

The invention provides a DMA (direct memory access) engine device and a data transmission method in a PCIE (peripheral component interface express)-to-RapidIO circuit, a DMA engine comprises a sending engine and a receiving engine which are used for processing data processing and protocol conversion of data packets sent and received in a DMA transmission process; the cache FIFO is used for caching the data read by the DMA engine; and the DMA channel control module and the state control register are used for controlling the DMA engine to update the operation state. A descriptor is configured in a PCIe memory through a host to define detailed parameters of DMA operation; the DMA engine can support DMA data transmission and protocol conversion between the PCIe core and the RapidIO core, and high-speed and low-delay data transmission in the PCIe-to-RapidIO circuit is achieved.

This invention discloses an intelligent early warning system and method for railway line patrol safety based on edge computing, involving intelligent safety early warning in the railway transportation field, enabling real-time controllable emergency measures to minimize the negative impact of incidents. The edge computing-based intelligent early warning system for railway line patrol safety includes: a cloud platform for an intelligent patrol system along the railway line, network infrastructure units for the intelligent patrol system along the railway line, on-site edge intelligent early warning devices and traffic control units, and IoT video sensing infrastructure units. This invention aims to deploy edge computers to continuously analyze 24-hour video streams from surveillance cameras at key locations along the railway line, process data in real-time, reduce latency, promptly identify abnormal events, take temporary emergency measures as quickly as possible, and rapidly report abnormal events, effectively shortening response time, significantly improving the efficiency of handling abnormal events in railway transportation, enhancing the accuracy of emergency measures, and ensuring railway line safety.

The invention relates to the technical field of disaster early warning processing, in particular to a mountain torrent disaster early warning system based on deep learning and air-space-ground integrated perception. The system comprises the following modules: a space-air-ground sensing module, an edge calculation preprocessing module, a deep learning prediction module and an intelligent decision and early warning module. Wherein the air-space-ground sensing module is used for collecting multi-source data; wherein the multi-source data comprises earth surface deformation data obtained through a spaceborne synthetic aperture radar, high-precision topographic data obtained through an unmanned aerial vehicle laser radar, hydrological sensing data obtained through a ground internet-of-things distributed sensor, and meteorological numerical forecasting data obtained through a meteorological data interface. Through air-space-ground integrated perception, physical constraint deep learning and edge intelligence technologies, mountain torrent multi-source data collaborative acquisition and accurate mountain torrent disaster prediction are realized, so that the comprehensiveness, reliability and timeliness of mountain torrent disaster early warning are improved.

The embodiment of the present invention provides a data packet sending method, device, electronic device and readable storage medium, which are applied to the field of computer technology. In the method, for any data packet to be sent, a target direction is selected from the available routing directions of the data packet to be sent according to a preset direction priority. In the case where there is no competing data packet to be sent in the target direction, the data packet to be sent is sent to the adjacent node in the target direction. In the case where there is a competing data packet to be sent in the target direction, routing arbitration is performed on the data packet to be sent and the competing data packet based on a preset arbitration rule. In the case where the arbitration result of the data packet to be sent is an arbitration failure, a target direction is reselected from the remaining available routing directions according to the direction priority, so as to send the data packet to be sent based on the reselected target direction. In this way, it is possible to avoid the data packet being blocked in the routing node for a long time, thereby reducing the data transmission delay in the entire network.

The invention discloses a GIS equipment wireless sensing monitoring system based on edge computing, and relates to the technical field of GIS equipment intelligent monitoring, the system collects the operation parameters of gas insulated switchgear in real time through an integrated sensor group and a monitor, preprocesses the data, locally computes and analyzes the preprocessed data, and obtains the real-time monitoring result of the GIS equipment. The method comprises the following steps: acquiring a gas quality index Qpz, a gas state dynamic index Qdz and an insulation state dynamic index Jzz, carrying out nondimensionalization on the indexes, calculating to obtain an equipment stability index SWDZ, comparing a preset threshold value with the dynamic indexes, providing primary and secondary evaluation, generating early warning information, and carrying out graphical display through an equipment operation state and an analysis result. Real-time data, historical trends and early warning information are provided, and a user is helped to identify problems and take measures; according to the system, real-time monitoring and fault early warning of the gas insulated switchgear can be realized, and the operation safety and reliability of the switchgear are improved.

The present invention relates to the technical field of cloud platform construction, and in particular to a method for constructing an intravenous treatment management cloud platform. The method comprises the following steps: performing gene expression pathway enrichment analysis on the patient's venous system through gene sequencing technology, thereby obtaining gene expression data; performing gene expression spectrum analysis on the gene expression data and performing intravenous treatment response gene labeling, thereby obtaining a treatment response gene labeling set; using cloud computing technology to build an elastic computing resource scheduling system, thereby obtaining resource scheduling system data; performing node load balancing weight scaling based on resource scheduling system data and gene expression data, thereby obtaining resource balancing scaling data; performing multi-scale pyramid feature refinement on gene expression data, thereby obtaining gene image feature refinement data. The present invention realizes the method for constructing an intravenous treatment management cloud platform through node load balancing weight scaling and GPU heterogeneous computing acceleration processing.